DDX3X Sits at the Crossroads of Liquid–Liquid and Prionoid Phase Transitions Arbitrating Life and Death Cell Fate Decisions in Stressed Cells

Several studies have shown the critical role of supramolecular structures that form membraneless compartments in regulating biological processes including cell fate decisions and cell signaling (Monks et al., 1998; Brangwynne et al., 2009; Goff and Lecuit, 2009; Franklin et al., 2014; Su et al., 2016; Banani et al., 2017). Some of the membraneless compartments are formed by liquid–liquid phase separation and can be readily disassembled by cellular machinery to restrict their function (Brangwynne et al., 2009; Case et al., 2019).

Stress granules are examples of such membraneless compartments and are assembled in the cytoplasm in response to some stressors to allow the cell to survive until the stressor is removed. Stress granules are highly dynamic structures containing mRNAs, 40S ribosomal subunits, and translation initiation factors, among other components (Kedersha et al., 2013; Anderson et al., 2015; Protter and Parker, 2016). Recent studies have highlighted the truly bewildering complexity of the biomolecules present in stress granules whose roles remain poorly understood (Jain et al., 2016; Khong et al., 2017). The composition of stress granules also varies based on the stressor that induces their assembly.

Liquid–liquid phase separation in stress granules is based on protein–protein, RNA–RNA, and protein–RNA interactions (Elbaum-Garfinkle et al., 2015; Molliex et al., 2015; Zhang et al., 2015; Mittag and Parker, 2018; Treeck et al., 2018; Van Treeck and Parker, 2018; Garcia-Jove Navarro et al., 2019). Defects in liquid–liquid phase separation can lead to an irreversible prionoid phase transition where a stable insoluble supramolecular assembly is formed. Amyloid fibrils formed in the brains of patients with Alzheimer's disease are a well-studied example of such assemblies (Patel et al., 2015; Ambadipudi et al., 2017). Defects in stress granule assembly machinery that shift the equilibrium of the process toward more stable prionoid phase transitions have been implicated in neurodegenerative diseases, partly by promoting aberrant programmed cell death (Dewey et al., 2012; Grad et al., 2015).

In addition to causing the assembly of stress granules, some stressors can induce activation of the NLRP3 inflammasome (Kesavardhana and Kanneganti, 2017). The NLRP3 inflammasome is a multiprotein heteromeric complex that contains the sensor NLRP3, the adaptor apoptosis-associated speck-like protein containing a caspase recruitment domain (ASC), and the catalytic actuator caspase-1 (CASP1). Activation of the NLRP3 inflammasome leads to: (1) the assembly of a cytoplasmic compartment, which is also likely membraneless, called the ASC speck, (2) CASP1-mediated proteolytic processing of the inflammasome-dependent cytokines IL-1β and IL-18 and of gasdermin D (GSDMD), and (3) the induction of a proinflammatory programmed cell death pathway called pyroptosis through cleaved GSDMD-mediated pore formation (Martinon et al., 2002; Kanneganti et al., 2006a, 2006b; Mariathasan et al., 2006; Kayagaki et al., 2015; Liu et al., 2016).

ASC speck formation involves a prionoid phase transition through homotypic caspase recruitment domain (CARD)–CARD and pyrin domain (PYD)–PYD interactions (Cai et al., 2014; Lu et al., 2014; Li et al., 2018). The ASC speck itself can induce pyroptosis in cells that take them up, providing an example of prionoid phase transition-driven programmed cell death (Franklin et al., 2014, 2018).

Stress granules and the NLRP3 inflammasome instigate two very different modes of supramolecular structure assembly in the cytoplasm of stressed cells, with stress granules driving liquid–liquid phase separation and the NLRP3 inflammasome promoting prionoid phase transitions. These processes provide two contrasting cell fate choices to the cell—live or die. However, the mechanism by which cells make this decision had not been elucidated.

In a study recently published in the journal Nature, we reported a mechanism through which murine bone marrow-derived macrophages (BMDMs) make such cell fate decisions (Samir et al., 2019). The central player in this decision-making process is a protein called DDX3X, which is a common essential factor for both the assembly of stress granules and the activation of the NLRP3 inflammasome. DDX3X is a DEAD-box family RNA-binding protein that had previously been found to be involved in stress granule assembly, translation initiation, Wnt signaling, and viral RNA sensing (Chuang et al., 1997; Yedavalli et al., 2004; Hilliker et al., 2011; Cruciat et al., 2013; Gringhuis et al., 2017). We observed that DDX3X is required for NLRP3 inflammasome activation as well.

The main objective of the study was to identify whether there is crosstalk between stress granules and the NLRP3 inflammasome. We chose to use sodium arsenite to induce stress granules and the combination of lipopolysaccharide plus nigericin to activate the NLRP3 inflammasome (Gurung et al., 2014; Aditi et al., 2018). We observed that prior induction of stress granules inhibited the NLRP3 inflammasome through an unknown mechanism. Using affinity purification mass spectrometry analysis, we identified DDX3X, a stress granule component, interacting with NLRP3.

Loss of DDX3X led to a decrease in NLRP3 inflammasome activation, suggesting that DDX3X itself is required for this process. Further analysis revealed that DDX3X is recruited to the ASC speck and is required for its assembly downstream of NLRP3 inflammasome activation. Thus, DDX3X is a common essential factor for stress granules and the NLRP3 inflammasome. Kinetic analysis using time course experiments revealed that stress granules and the NLRP3 inflammasome compete for DDX3X molecules. This competition allows the cell to choose between life and death based on the subcellular localization of DDX3X molecules, with DDX3X in stress granules being prosurvival and DDX3X associated with the NLRP3 inflammasome being pro-pyroptosis (Fig. 1).

OPEN IN VIEWER

FIG. 1.

Subcellular localization of DDX3X allows the cell to make life and death decisions. (A) DDX3X localization in stress granules promotes stress granule assembly and cell survival. (B) DDX3X association with the NLRP3 inflammasome promotes a type of proinflammatory programmed cell death called pyroptosis.

This study suggests a competition-based mechanistic paradigm for the regulation of the inflammasome based on DDX3X cellular valency (Samir and Kanneganti, 2019; Samir et al., 2019). At the same time, it has raised many interesting questions that need to be addressed. One of the key questions relates to how DDX3X molecules decide whether to go to stress granules or the NLRP3 inflammasome. Coimmunopurification experiments suggest that DDX3X and NLRP3 physically associate before inflammasome activation.

Kinetic analysis suggests that although prior induction of stress granule assembly inhibits the NLRP3 inflammasome, simultaneous induction of stress granule assembly and NLRP3 inflammasome activation reduces CASP1 cleavage by approximately half compared with the amount of CASP1 cleavage after induction of NLRP3 inflammasome activation alone. Induction of stress granules after the inflammasome activation signal had been provided did not have an effect. Therefore, DDX3X subcellular localization seems to depend on which signal is provided first. This might be dependent on stochastic factors in the case of simultaneous application of stress granule-inducing and NLRP3 inflammasome-activating signals. However, this does not necessarily rule out more explicit mechanisms governing DDX3X recruitment to the NLRP3 inflammasome.

Not all stressors, for example, osmotic shock and thapsigargin treatment, were able to induce assembly of stress granules in LPS-primed BMDMs, while they could induce this assembly in other cell types. This might be due to cell type-specific differences in physiology of BMDMs compared with transformed cell lines. The exact nature of the differences and the underlying mechanisms generating these differences will need to be explored further.

Reversibility of liquid–liquid phase separation can allow a cell to dynamically regulate its physiology by sequestering, releasing, and/or increasing local concentrations of biomolecules. A recent report showed that the ATP-bound forms of DEAD-box proteins such as DDX3X promote phase separation of ribonucleoprotein complexes into membraneless subcellular compartments with liquid-like properties, such as stress granules (Hondele et al., 2019). ATP hydrolysis reverses this process, inducing the disassembly of the membraneless compartments. Based on published studies, one could hypothesize that liquid–liquid phase separation, at least in initial stages, generally serves to promote cell survival.

It is interesting to note that the helicase activity of DEAD-box proteins such as DDX3X is typically required for protein function (Rocak and Linder, 2004; Hilliker et al., 2011; Mugler et al., 2016; Hondele et al., 2019). Inhibition of DDX3X helicase activity promotes assembly of stress granules in the murine L929 cell line but fails to induce stress granules in BMDMs (Samir et al., 2019). In addition, DDX3X helicase activity is dispensable for NLRP3 inflammasome activation, suggesting that DDX3X has a scaffold function.

Several proteins involved in innate immune signaling and regulation seem to have distinct catalytic and scaffold functions. This includes receptor-interacting serine/threonine protein kinase 1 (RIPK1) and caspase-8 (CASP8), which play a central role in executing programmed cell death including pyroptosis, apoptosis, necroptosis, or PANoptosis. PANoptosis is a recently described programmed cell death pathway induced by assembly of a heteromultimeric protein complex, the PANoptosome (Christgen et al., 2020; Samir et al., 2020; Zheng et al., 2020; Malireddi et al., 2019). RIPK1 and CASP8 are components of the PANoptosome (Christgen et al., 2020; Zheng et al., 2020), and their scaffold function is critical for the execution of cell death (Fritsch et al., 2019; Malireddi et al., 2019, 2020; Newton et al., 2019a, 2019b; Lalaoui et al., 2020; Tao et al., 2020). The scaffold functions act to keep multiple valences of proteins together, and multivalency is critical for assembly of these supramolecular structures (Samir and Kanneganti, 2019).

The DDX3X scaffold function seems to drive assembly of ASC specks after inflammasome activation. ASC specks have prion-like architecture and cannot be easily disassembled. Therefore, it can be argued that the catalytic activity of DDX3X modulates a dynamic liquid–liquid phase separation for stress granules, while its scaffold function regulates a stable prionoid phase transition for the NLRP3 inflammasome.

The exact molecular mechanism through which DDX3X regulates and accomplishes these two very different types of processes remains to be elucidated. Furthermore, the role of these phase transitions in additional processes should be determined. For example, execution of PANoptosis requires assembly of a novel cytoplasmic compartment containing RIPK1, CASP8, and ASC (Malireddi et al., 2020). Assembly of this PANoptosome occurs in the absence of transforming growth factor beta-activating kinase 1 (TAK1). Since ASC is capable of prionoid phase transitions, it is tempting to speculate that the assembly of the PANoptosome also involves prionoid phase transitions. Future experiments will be needed to test whether assembly of the PANoptosome indeed involves a prionoid phase transition and to unravel the molecular mechanisms controlling this process.

Mutations in DDX3X are associated with cancers and a recently described hereditary neurological disorder called DDX3X syndrome (Snijders Blok et al., 2015; Valentin-Vega et al., 2016; Wang et al., 2018). Earlier studies implicated ASC prionoids in neurodegenerative diseases (Franklin et al., 2014). Since the scaffold function of DDX3X can promote the prionoid phase transition of ASC, it is possible that DDX3X syndrome-associated mutations in DDX3X are also promoting similar phase transitions. It would be informative to test and compare the ability of the wild type and mutant DDX3X proteins to induce or undergo prionoid phase transitions.

In summary, observation of crosstalk between the cellular stress response and inflammasome pathways and elucidation of a molecular mechanism governing this crosstalk have provided us with a novel targetable process for therapeutic interventions in human diseases. We hope that future research will exploit this knowledge to devise precise and effective therapies for DDX3X syndrome and cancers involving mutations in DDX3X.